While improvements in living standards and medical technology have increased the elderly population in Korea, there is also an increased incidence of chronic diseases. Thus, the demand for medical services has increased rapidly and the demand for nurses has also increased consistently. The nursing shortage is a serious problem in Korea, which has the lowest ratio of nurse active in the workforce per 1000 people of all 19 OECD countries (Kim, 2010). However, hospitals are not actively hiring more nurses despite the increasing patient numbers. The job satisfaction of the frontline nurses is much lower than that of other hospital professionals because of the irrational shift system (Kim et al., 2010). Furthermore, the workload of nurses has increased recently because of severe competition among hospitals, and the introduction of an evaluation system for medical institutions. Over the past 10 years, the turnover rate of nurses has been 15-20% (Kim). Turnover among nonsupervisory registered nurses comprised over 90% of nurses turnover, and more than 75% of the leaving nurses moved to nonclinical positions outside hospitals, which has resulted in a loss of medical human resources (Kim).

The turnover of hospital nurses is closely related to job stress (Meadors & Lamson, 2008; Yom, Ko, Kwon, Kwon, & Lee, 2009), which has many causes (Kim et al., 2010). In the present study, we investigated job stress among hospital nurses from the perspective of compassion fatigue. The concept of compassion fatigue was first introduced by Joinson (1992) to describe incidents where nurses collapsed because of an excessive workload in hospitals. Joinson studied the compassionate and sympathetic relationships between nurses and patients that resulted in the burnout of nurses, which was defined as compassion fatigue. Compassion fatigue is a stress caused by the activity of helping other people in need, or those experiencing trauma or pain, and it is a natural consequence of helping people who are suffering extreme stress (Figley, 1995). Professionals such as nurses, doctors, social workers, fire fighters and police officers are repeatedly exposed to work-related traumatic events and their risk of developing posttraumatic stress disorder (PTSD) is high (Figley, 1995). Compassion fatigue was initially referred to as secondary traumatic stress, but this terminology may have attached a stigma to people suffering mental impairment, so the term compassion fatigue is now used widely (Figley, 1995, 2002). People with compassion fatigue are insensitive to fatigue, emotional lethargy, social withdrawal, and the emotions of passion and hope, and they experience physical and emotional problems, as well as problems with human relationships (Saakvitne, Gamble, Pearlman, & Lev, 2000). The symptoms of depression, anxiety, impulsive behavior, substance abuse and somatization, and burnout develop if compassion fatigue is cumulative (Yoder, 2010). The state of chronic burnout is detrimental to the personal lives of sufferers and it adversely affects the job performance of individuals and groups (Collins & Long, 2003; Meadors & Lamson, 2008). It may also be a factor that affects turnover (Collins & Long).

Previous studies of compassion fatigue in nurses include the following : a study in the nursing environment that used a hybrid qualitative and quantitative approach to understand the relationships between compassion fatigue, medication errors, and ethical conflicts among nurses (Maiden, 2009); a study of recognition, emotions, and experiences of emergency room nurses who helped hurricane victims (Frank & Adkinson, 2007); a study of a professional group that included doctors, nurses, behavioral scientists, and trauma workers who worked in a children's hospital in Philadelphia, USA (Meadors & Lamson, 2008); and a study of the prevalence of secondary traumatic stress in emergency room nurses (Dominguez-Gomez, 2009).

Previous studies of compassion fatigue or secondary traumatic stress in Korea include: a study of compassion fatigue and burnout in fire fighters (Choi, 2009); and a study of the recognition of compassion fatigue, burnout and compassion satisfaction among counselors working for child protection institutes (Shin, 2007). Most of the studies conducted since 2000 have been other than nursing science. A recent study of the relationships between compassion fatigue, burnout and compassion satisfaction in emergency room nurses was conducted by nursing scientists (Kim, 2011), but few other study results have been reported.

By contrast, studies of turnover intention, i.e., the intention to leave a job in the near future, which is recognized as a cognitive antecedent (Hayes et al., 2006) are relatively common because turnover intention is used as a variable to predict turnover (Jeong, Kim, & Kim, 2008; Park, 2006). Previous studies of turnover intention among nurses have shown that job stress and job satisfaction are important factors that affect turnover intention (Parry, 2008; Yom et al., 2009). Various studies have been conducted in Korea and, elsewhere to determine the factors that affect turnover, but the high turnover rate of nurses has not been. Previous studies have focused on job stress, which is considered to be a factor that affects turnover intention. By contrast, the relationship between turnover intention and compassion fatigue was investigated in the present study.

The investigation of compassion fatigue level is very important because of the influence of compassion fatigue on human performance and burnout (Shin, 2007). Untreated compassion fatigue can lead to burnout, so it is necessary to determine the level of compassion fatigue. The relationship between compassion fatigue and turnover intention among hospital nurses also needs to be understood.

From the perspective of compassion fatigue experienced by hospital nurses, the present study investigated relationships between compassion fatigue as a predictive variable for turnover intention.

The purpose of the present study was to investigate the level of compassion fatigue in hospital nurses, and the relationships between compassion fatigue, burnout and turnover intention. Our detailed aims were as follows:

First, to investigate the levels of compassion fatigue, burnout and turnover intention in hospital nurses.

Second, to investigate the level of compassion fatigue according to general working conditions of hospital nurses.

Third, to investigate the relationships between compassion fatigue, burnout and turnover intention in hospital nurses.

In the present study, we investigated the levels of compassion fatigue among hospital nurses, and tested for correlations between compassion fatigue, burnout, and turnover intention. The mean level of compassion fatigue in the subjects was 50.58, which was 'very high' according to the evaluation standard produced by Figley and Stamm (1996). This result was similar to those reported by Mangoulia, Alevizopoulos, Fildissis, Katostaras, and Koukia (2011) who detected a high level of compassion fatigue in ICU nurses, and by Shin (2007), who reported a high level of compassion fatigue in more than half of the counselors in child protection institutions. By contrast, Frank and Karioth (2006) did not detect a high level of compassion fatigue in their study of nurses who cared for hurricane victims. The results of the present study were also different from those reported by Kim (2011) who detected 'neutral' compassion fatigue in emergency room nurses. The present study included 23 questions where the scores ranged from 0 to 5, whereas Kim used 10 questions with a 5 score scale. The reason of discrepancy between the results of the present study and that of Frank and Karioth may have been because the nurses who participated in the latter study were aged ≥40, and they worked for a short period of 2 weeks with hurricane victims. In the studies of Hooper, Craig, Janvrin, Wetsel and Reimels (2010), the level of compassion fatigue in emergency room nurses was higher than that of nurses working in other departments, although the difference was not statistically significant. However, Kim detected a lower level of compassion fatigue in emergency room nurses compared with the present study. Thus, the level of compassion fatigue in nurses may be related to the work department or the patients.

In the present study, the tests of the relationships between the compassion fatigue scores and the of nurses detected high scores in young nurses aged 20-29 years, nurses with <3 years of total length of service, nurses with <2 years of current service in their department, and nurses who were not satisfied with their remuneration. According to the study of ICU nurses by Mangoulia et al. (2011), nurses who were female, married, and who had master's degrees had high compassion fatigue scores. According to the study of counselors in child protection institutions conducted by Shin (2007), female and single subjects had higher compassion fatigue scores. A study of nurses who cared for cancer patients, which was conducted by Potter et al. (2010), reported higher compassion fatigue scores for nurses who cared for inpatients, nurses with 11-20 years' service or service in the current department, and nurses aged 36-50 years, although theses results did not agree with the present study. The result of present study showed that nurses with a shorter duration of service experienced more compassion fatigue, which agreed with studies of middle-aged nurses conducted by Frank and Adkinson (2007) and Kim et al. (2010). The overall duration of service is a variable that affects compassion fatigue (Potter et al.), and it also affects the levels of other experiences. Highly experienced nurses cope with critical situations or trauma stress using various methods (Potter et al.). A shorter overall duration of service or duration of current service in a department, increases the pressure nurses experience at work, their stress and levels, and compassion fatigue, and they may have difficulty adapting to situations (Kim et al.). The level of experience may affect ability to cope with difficulties, so emotional support programs are required for less experienced nurses. Initially, there should be opportunities for sharing indirect experiences with co-workers and senior nurses in the same department. The ability to cope with compassion fatigue may improve if an intervention is provided in the form of an emotional support program.

In terms of the relationship between compassion fatigue and burnout, most of the studies including the present study, have reported that higher compassion fatigue scores are associated with the higher level of burnout (Frank & Karioth, 2006; Hooper et al., 2010; Kim et al., 2010; Shin, 2007). Compassion fatigue is a variable that affects burnout (Hooper et al.; Joinson, 1992), and which is induced by caring for patients who have experienced traumatic events (Figley, 2002; Potter et al., 2010). Burnout is a form of physical, emotional and mental exhaustion that results from unrealistic and excessive demands (Maslach & Jackson, 1981). Nurses with burnout experience various physical symptoms and lethargy, and they become indifferent to their patients (Kim et al.). Burnout and compassion fatigue are similar in terms of the symptoms of lethargy, despair, anxiety and depression (Lee & Yu, 2010). Joinson described compassion fatigue for the first time and explained that it affects professionals who care for people, such as nurses, psychologists and pastors. However, while burnout results from chronic and excessive job-related stress (Yoder, 2010), compassion fatigue may result from even a single exposure to posttraumatic stress (Figley). Eventually, prolonged compassion fatigue can lead to burnout, which confirms the close relationship between them.

Nurses experience traumatic events while caring for patients and compassion stress arises because of factors such as loss of control, a feeling of being overwhelmed, frustration and adverse environments (Kim et al., 2010). Compassion fatigue arises if compassion stress is not properly managed and emotional, physical, and emotional deviations may develop in the worst cases. If nurses continue to work under these psychological and emotional burdens, they experience burnout with physical and mental exhaustion (Shin, 2007). In particular, job-related burnout can reduce the productivity and efficiency of teams and a nurse may eventually leave the team (Collins & Long, 2003). Various intervention modalities should be explored to find solutions. For example, Choi (2009) suggested that fire fighters should be provided with appropriate remuneration to reduce their levels of compassion fatigue and burnout. Compassion fatigue and burnout may increase if nurses who have experienced compassion fatigue or burnout allocated to the same environment. Therefore, employment placement is important for nurses (Frank & Karioth, 2006). In addition, the levels of compassion fatigue and burnout should be used as an evaluation tool for the employment placement of hospital nurses, because the employment placement may accelerate the turnover intention of nurses (Frank & Karioth).

In the present study, hierarchical regression analysis was conducted only variables for compassion fatigue and burnout variables, whereas we ignored other control variables in order to investigate whether compassion fatigue and burnout were predictive variables for turnover intention. Compassion fatigue accounted for 29.6% of the variance for turnover intention, while compassion fatigue and burnout accounted for 42.2% of the variance.

Changes in the working environment caused by the turnover of nurses can increase the workload and stress for the remaining nurses. This affects the anxiety levels of patients because the quantity and quality of nursing services are reduced and the safety of patients is also affected adversely. The hospital incurs additional costs recruiting and training nurses (Liou, 2009). The turnover of nurses is an adverse factor that affects medical environments which are required to provide good quality of nursing services (Liou). Turnover intention is recognized as a cognitive antecedent (Hayes et al., 2006), and it has been used widely for predicting turnover in many studies in Korea and elsewhere (Jeong et al., 2008; Park, 2006). The turnover intention of nurses is associated with job stress and job dissatisfaction (Parry, 2008; Yom et al., 2009) and burnout (Kim et al., 2010).

Individual efforts are required to control adverse factors in order to reduce compassion fatigue among hospital nurses. There should be opportunities to produce an increased sense of satisfaction such as exercise, leisure activities, sleep, proper nutrition, and spending time with loved ones, and psychotherapy could be used if necessary (Badger, 2001). From an organizational point of view, stress is accelerated by compassion fatigue, so opportunities should be provided for taking a break, while nurses experiencing high tension and stress should be observed and supported (Potter et al., 2010). Compassion fatigue in hospital nurses is associated with burnout and turnover intention, so a strategy should be established to reduce the levels of compassion fatigue and burnout in hospital nurses, which may lower the turnover intention.

A limitations of the present study was that the samples were not collected and randomly so any generalization of the results is limited. Control variables were not included in the hierarchical regression analysis. Future work needs random sampling, comparison and analysis of nurses in different work departments, and human resources management to reduce compassion fatigue, burnout, and turnover intention.

This study showed that the level of compassion fatigue in hospital nurses was high, particularly in young nurses with short length of service. Therefore it is necessary to reduce compassion fatigue in young nurses, which may also reduce the turnover intention among Korean hospital nurses.